NOTE - files.eric.ed.gov
Transcript of NOTE - files.eric.ed.gov
DOCUMENT RESUME
ED 313 696 CS 212 023
AUTHOR Olson, David R.TITLE Interpretation and the Autonomy of Written Texts.PUB DATE Mar 89NOTE 21p.; Paper presented at the Annual Meeting cf the
American Educational Research Association (SanFrancisco, CA, March 27-31, 1969).
PUB TYPE Speeches/Conference Papers (150) Reports -Evaluative /Feasibility (142)
EDRS PRICE MF01/PC01 Plus Postage.DESCRIPTORS Ambiguity; *Cognitive Development; Educational
Research; Elementary Education; Memory; TextStructure
IDENTIFIERS Discourse; *Text Factors; Text Learning
ABSTRACT
Textual meaning is not autonomous--not only do themeanings of texts change as contexts change but also the textual orsentence meanings change as cultural conventions change. A series ofpreliminary studies have established that children into the early-chool years believe that the speaker's intentions, especially when_he speaker's belief is false, make their way to the listener evenwhen there is no lexical means for conveying that intention. Thesestudies indicate tat children tend to conflate what a speaker meanswith what the utterance means. With age and schooling they come tosee there two "meanings" as independent. Children come to revisetheir estimate of the meaning intention on the basis of thelinguistic or sentence meatinj. They begin to treat that textualmeaning as autonomous. But what continues to be the case is that thebasic di-;tinctions required for literate interpretation continue tobe useful for distinguishing between the properties of the text whicnare "taken as given" for any particular purpose and the set ofconstruals or interpretations that can be made of that text. Inlearning to distinguish meanings from intentions, and thoughts fromexpressions, children take the first step on the path which proceedsfrom taking facts as given and then organizing those facts into sometheoretical scheme. (Thirty-eight references and a table of data areattached.) (RS)
***********************X*****x****************************************** Reproductions su,plied by EDh are the best tat can be male ,t
* from the original docillaent. ,t
*******************************************A****************A**********
"PERMISSION TO REPRODUCE THISMATERIAL HAS BEEN GRANTED BY
Darlct F. Ciaoyi,
U I DEPARTMENT OF EDULATIONOffice of Educational Research and Improvement
EDUCATIONALCENTER IRESOURCE CONFORMATION
This document has been reproduced esERI
received from the person or organizaticoiginating itMinor changes -lave been made io improvereproduction dualif
te0TO THE EDUCATIONAL RESOURCES D_OlsonINFORMATION CENTER (ERIC)''
i-oiros vi.meatdo not necessarily represent officio,
OERt pOSitOr, or PChCy
Interpretation and the autonomy of writtentexts
CteJ
r-i David R. Olson
Ontario Institute for Studies in Education. Tot onto.Canada
rix4 The orality literacy hypothesis as developed by Havelock (1982). Goody and Watt
(1963), McLun-.n (1962) and others (for a summary of recent research under that theme
see Olson, 1987) has encouraged a new and promising approach to the problem of the
meaning and interpretation of language and text. The argument, in part, is that
writ -en discourse invites a new form rf producing and interpreting language because it.
unlike oral discourse, is typically separated from its author in time and space. A book, a
paper or a memo written in Toronto on March 15 (beware the Ides) may be iead by
someone in Missoula on April 15. Indexicals like "today" and "here" used by the writer
will not be shared by the reader. Interpretation, on the readers part will require a
reconstruction of the reference of such indexical expressions. We may think of this
problem in terms of "de,-;ontextualization" and "recontextualization" of written texts.
Writers. realizing the autonomy of their texts will either explicate or minimize the
unsharable aspects of cpntext in such a way that the appropriate inter pi etation is more
readily reconstructed by the reader. Such procedures help to create a kind of text i-hich
is radically different so the argument goes, from more interactive and contextualized
forms of discourse.
AERA89 ms,-;
Recent writers have pointed Out that sucn autonomous texts are not unique to
liteiate traditions. Narasimhan (in press) points to the fixity of traditional Indian Vedic
poetry. Since 600 BC a single standard version of the Rgveda has been preserved and
transmitted from generation to generation. The fixity of form is guaranteed by
'Paper prepared fol the Symposium ''The autonotry of text at the 1989 Conference of the AERA bar.Francisco. March 1989
nJ
ingeneous mnemotechnics which involve various transformations on the text. treating
separate words, separate consonants and vowels and the Like. Consequently. oral
performances of these Vedas in different parts of India are all identical. All without the
use of writing as a device to fix text and to serve as a standard for production. Goody
(1987. p. 122.) however. has argued that the Vedas 'bear all the hallmarks of a literate
culture." the suggestion being that the mnemonics invented appear to be based on a
written transcript rather than on a memorized text.
Feldman en press) too, has pointed out that ordinary oral discourse has devices
for fixing text in such a way that it can be referred to in subsequent discourse. direct
quotation being - ileac and simple example. She cites numerous cases of discourse in
traditional societies in which the speech of one person becomes the object of a second
person's speech. Turning .:,peeeh into an object of discourse is one means of "freezing" a
text. Just what is frozen. the wording or the semantic content or some combination,
remains to be seen.
Yet the properties of such "autonomous" texts differ in important ways depending
on whether they are created under "oral" conditions or under "literate" conditions.
Havel'-k, building on the original work of Milman Pert:. showed how the constraints of
human memory shaped oral texts for memorability by recounting the actions and
experiences of gods and heroes, and expressing them in a language of formulaic phrases
and poetized speech. Havelock was careful to point out that such oral tradition is
anything but simple. The Homeric Greek tradition involved a complex literature and a
complex social or '4er. But it was framed. as he said. on the memorized word. Goody and
Watt (1963) made a similar argument based on their study of contemporary oral myth.
Story telling, also a part of an oral culture, does not involve the construction of
"autonomous" texts. Stories have a fixed content but not a fixed verbal for m. To be a
text in the sense I am using the term requires a fixed verbal formula: the text is fixed. if
you like, at tha level of surface structure. Ritual ,zpeech and poetized speech tend to fit
J
:3
the defir_ition although both Finnegan (19S ) and Goody (1987) have pointed out that
even ritualized speech in the oral societies they ,,tudied retain a certain degree of
variability in performance: the same ritual is rarely performed using "the same words"
Indeed. the very concept of the same words appears not to be .ranslatable from oral to
literate culture.
Thus, while there are both oral and written texts, the structure of such texts
differ importantly, the one being based on the properties of oral, verbal memory the
other based on recoverability and readability. Indeed, Luria (1976) suggested that with
the rise of literacy the organization of memory shifts from associative systems to the
categorial systems implicated in definitions, lists and other literate artifacts.
A second contrast between oral and written texts is the particular conception of
meaning that each give rise to. In my first attempt at this question (Olson, 1977) I
suggested that in oral discourse. the meaning is to be found in the intentions of the
speaker whereas in our Western literate tradition, meaning is to be found in the
autonomous text quite independently of the intentions of the speaker. This has raised a
good deal of controversy and called for some revision. Two basic problems were raised.
One is that the distinction is not simply an oral/written one because there are literate
cultures which do not treat texts as autonomous, the Moslem culture apparently being
one (Scribner and Cole, 1981; Streec, 1984). Rather, the claim has to be that in Western
literate culture, texts have been exploited in such a way as to institutionalize the idea of
the autonomy of texts. In a word, in the dominant Western tradition. texts have come
to be treated as objective and autonomous embodiments of meaning. -Indeed, a part of
the history of the Western tradition could be described as the systematic development of
a form of discourse and a species of symbols that could be autonomous. essayist prose.
mathematical expressions and computer programs being extreme cases. Perhaps the
most renowned exposition of this view is Sir Karl Popper s theory of "objective
knowledge" (Popper. 1972 ) Such knowledge P-; tied to or exploits literacy but, of course ,
literacy did not cause it to happen
4
The second problem is that it has been argued by Rommetveit (1988). Nystrand
(1986) and others that the whole notion of textual meaning is a mistake. Rommetveit
quotes approvingly from Volosinov who says "The fiction of a words realm promotes
the reification of its meaning" (p. 23.) That is. the fact that a written or punted word
appears to be substantial-- it is permanant in time and space. can be put in one's pocket
and so on-- has led us to a misplaced concreteness. We must remember. they point out.
that a written word is just a mark on a paper It has no intrinsic meaning-- it is a
convention we have agreed to use "to express a shared meaning". Meaning is.
Rommetveit argues. "intertextual," not storable in an object but existing as an
agreement between persons. Nystrand argues in a similar way that meaning is between
speakers and listeners; language has meaning only because it specifies a relation
between speakers and listeners or readers and write' s.
Consequently. the tendency prevalent among most literate people to think that a
printed mark actually "contains" meaning is a simple error. We are duped, they say. by
the concreteness and permarre of the mark. In one sense, [ agree we are often seduced
by th- phenomenal experience of suddenly grasping the meaning of a text into believing
that the meaning was there to be grasped all along. Consequently, we as teachers.
especially as teachers of beginning readers, may mistakenly believe that if a student
fads to recover that meaning, the problem could be remedied by a closer look at the text.
Meaning, we sometimes believe, can be recovered by excavation. Indeed, Luther thought
so, claiming that the meaning of scripturo required not the dogmas of the church but "a
deeper reading of the text" (Gadarnar, 1975.)
We now acknowledge that Luther was only telling half of the story. One can look
at a text forever without seeing its "preferred" meaning. Reading, like observation in
science, requires a prepared mind Boring (1950) pointed out what biologists "saw"
when they looked through a microscope at a cell before and after the discovery of Lhe
microscope; before, they saw granulated matter. after. tney saw chromosomes. Seeing
meaning in a text requires the availability of the appropriate set of concepts just as
5
much as observation in science does. Recent reading theor, has done much to drive this
point home (Smith. 1971: Anderson. 1977). Reading is not so much recovery as it Ls
recognition.
But Luther was right. I would agree. about the fact that reading is recognition of
structures which are actually there in the text. Contrary to much of criodern reading
theory. reading is not guesswork nor is it invention or fabrication or assimilation or
hypothes;s testing; or problem solving; it Ls recognition. For an analogy consider the
famous Gestalt drawings with hidden figures: those figures are actually there.
Discovery in science is not sheer invention: it is seeing nature through or in terms of
complex theory. Hansen's (1958) claim that observation is theory-laden means that one
needs the theory to see w}.-at is there but also that the observation is genuinely the
detection of structure.
So I am arguing that the concreteness naively attributed to the written word is
not entirely wrong; it is just naive. Structure in a word or a text is in the text but it is
detectable only by one who looks at the text with the appropriate hackground knowledge.
Nor are interactive theories of reading correct on this point. Reading is not an
interaction between data driven or bottom-up processes and concept driven or top-down
processes. The lowest level structures such as the letter "c" can be recognized only by
someone with the requite knowledge. Similarly, the highest level structures. say
metaphor, can be recognized only by someone with the requisite knowledge. Both are
equally dependent upon the prepared mind and both are equally given in the text. It is
a false dichotomy to think that some things are in the world and some are in the mind.
The mind, as Gibson (1966) repeatedly argued, is best thought of as attunement to the
invariants in the world.
But what Ls this knowledge? It includes such things a:: knowledge about the uses
of texts the culture. the issues that a text can adjudicate. birth certificates and land
titles for example, and knowledge of who has the authority to interpret a text, priest.
6
judge or oneself. Vernacular literacy, it may be noted, is important for the access it
gives people to the documents and texts that control them. It also includes knowledge
about orthography and types of discourse. But it also, and this may be more
controversial, it includes knowledge of the meanings of words and sentences
independently of the uses w which those meanings are put for the purposes of the
speaker or writer. This knowledge may be referred to as "word meaning" or "sentence
meaning" and it may be contrasted with the more general "speaker's meaning" or
"intended meaning." It is to this knowledge that we now turn.
First, it may be worth acknowledging that knowledge of the meaning of words
separate from the use of the word or sentence in expressing a speaker's intention. may
not be strictly tied to literacy. Some linguists halve argued that understanding a
language requires that the speaker possess this knowledge even if the knowledge remains
implicit, that is, unavailable as an object of thought or discourse. My suggestion is that
literacy, at least Western literacy, has placed a high premium on this knowledge and
,as made much of it explicit. We have dictionaries, schools, cross-word puzzles, spelling
lists, thesauras and SATs. Further. available evidence (Harriman, 1986. Francis. 1988)
suggests that such knowledge is acquired when children are five or six years of age
when they first encounter print. Finally, it appears that the acquisition of that
knowledge has important cognitive implication:. To hurt at just one, it is knowledge or
awareness of "sentence meaning" which permits writers to revise their texts to bring the
sentence meaning into congruity with their intended meaning; conversely, it is that
knowledge which permits readers to re-read a text with altered understanding.
Let us now consider chridt en's acquisition of an und, -standing of the relations
between what sentences (or words) mean and wi.at they, as speaker's, mean by them
Note that the claim is not simply that one is in the text and the other in the world.
They are both represented by in the text ii' a way analogous to that in which both the
letters and the words are in the te.st. The questions are, when does the child detect
these two levels of structure and recognize them as distinctiN e proper ties, and what role
r
7
does literature play in this development?
Texts as autonomous representations of form
When they are very young children 1,egin to treat texts as fixed verbal for
Parents frequently report that children will "correct" their reading of a text if it errs in
any way from the child's memory of the text. Nancy Thtrance, Elizabeth Lee and I have
recently documented this observation by systematically inserting "misreadings" into the
reading of a well known story. Sixteen children. 2-6 to 3-9 years of age, heard a story
six times and on the seventh reading the reader introduced a change in the text:
policemen became fireman, brothers and sisters became cousins, and a "clang" became a
"crash."
Six of the children showed no sign of awareness of any of these changes. However,
10 of them interrupted the reading, insisting on the correct text. Indeed, another child,
not one of the sixteen. insisted that the mother had incorrectly read the text when, in
fact, she had read precisely what it said. Presumabiy it deviated from the child's
memory of the text.
Such fIndings indicate that stories read to children play into their "oral memory"
to create notions of fixity of texts. Such written texts then are somewhat analagous to
remembered songs, nursery rhymes, jingles with this one difference rhe remembered
story has its external representation, the marks on the paper in the text.
This understanding is shown in another way. Tom Keenan, Nancy Torrance and I
designed simple cartoon drawings with a line of text under each of the drawings. One
picture, for example, showed a drawing of Charlie Mown, and under the picture was
written the work "Charlie." The picture and the text were identified for the children.
Children were then asked a series of questions about the picture: Is this a picture of
Charlie Brown? Is this a picture of a little boy' Chen they were asked the following
about the text: Does this say "Charlie Btown" Des this say "a little boy?" By the
time they were three, most children realized that pictures may be described in mote than
C;
S
one way but that a text even a text they themselves cannot road,
one way. The text has a fixed reading.
ran he :n only
In their early encounters with fixed texts children are learning an important fact
about texts, namely. their autonomy. Their form and meaning are not tied to the
occasion of utterance or to the speaker/reader. They are texts. pieces of language. which
are objects of recitation, remembering and understanding It is a curious fact that
children are so vulnerability to these texts. and to all early soi.gs and stories, forms or
lang'iagc which, at least at the beginning, lack any meaning or significance. Perhaps
Havelock .s right in suggesting that we are evolutionarily shaped for attending to and
storing away oral texts.
Autonomy of meaning
For a number of years my colleagues and I have been attempting to understand
what children take to be the meaning of an utterance or a text. Do they think the
meaning is the thing the text refers to'? Do they think the meaning is what the c;peaker
or writer intended by the text? Do they grant a certain autonomy to the meaning of a
text distinguishing what a text means from what a speaker means by it?
This, of course, is not a straightforward question as it is the very question against
which adult philosophers bang their heads. I believe it is fair to say that the three.
historically orderea. theories of meaning that are most commonly held are what we may
call the "Intentionalist theory." the "Literalist theory." sometimes referred to as the New
(new in the 30s) Critisism and the "Reader response theory." My interest here is not in
exposing the flaws in these theorie; but rather in attempting to understand just what
children's theories of interpretation are and how they change with the development of
literate competence.
Our primary hypothesis has been. and continues to be until we get a better idea.
that younger children conflate saying with meaning. Indeed. such conflation should be
e: ,,'?cted in view of the fact that in our vernacular (I could say "oral") speech, the verb
say has two senses. What one said could be reported by direct quotation or indir >et
quotation. Ask and tell are similar in this regard: If someone asked for a glass of wv fel
we may report this either directly by saying: He said 'Could I have a glass of water er
indirectly as his asking for a drink. If someone told us "Get out" we could report this as
having been told where to go. and so on.
But it is also important to notice that when we are speaking more "precisely" (I
could say "literate"), we would hedge the reports, saying "He didn't actually say" or add
"in so many words" and she like to show that we do often distinguish what a speaker
actually said, the very words. from what the speaker intended to communicate.
Furthermore, we do have lexical items for distinguishing what was said. from what was
meant, from what was intended, from how it was interpreted and the like. Control of
such distinctions provides a great deal of latitute for talking and thinking about what
people mean when they say something (See Olson and Astington, 19S9 for a discussion of
this issue.)
There is an equally signifIcant piece of the more technical (I could say "literate")
lexicon that bears directly on the wording of the text, the locution as Austin (1965) and
Feldman (in press) refer to it, that we use to distinguish the meaning of a word or
sentence from its use on a particular occasion. When we refer to the meaning of the
word "apple" we are discoursing on the word as an autonomous object, quite
independently from the particular apples in the fruit bowl in front of us. Some have
argued that meaning never comes to he free of reference as did I several years ago
(Olson, 1970) but that view now seems to me to be impossible. Some distinction has to
be made between the 'sense' of the term and what that term was used to refer to in a
particular case (Frege, 1952). Words have analyzable meanings which are just as
objective as anything else we have in our socially conventionalized world (cf. human
rights.) Whether these objective properties extend to larger units of discourse remains to
be seen. The meaning of a poem does seem to be less objective than the meaning of a
term.
1u
10
Even the meaning of a term may not be completely independent of context.
Rommetveit (1988) has criticized my view of the "autonomy of meaning" of terms by
reexamirig one task we have used si,ith children. a story that was first studied by
Newman (1982). In this story. Earnie. a Sesame street character. tells his credulous
friend Bert "I'm going to divide this banana up so both of us can have some" ,,hereupon
I . eats the whole banana and gives the skin to Bert. I had argued that in the early
school years. children come to realize that "pa7t" has a literal meaning and that.
therefore. they would acknowledge that Earnie 'iad not merely lied. 'His action, literally
(I could say "literately") followed from his words. Children have come to understand, I
claimed, the autonomy of texts.
Rommetveit argued. on the other hand. that the meaning of "part" is context
dependent. Part in tne discourse on sharing food literally means an edible part; part in
biology class means any discernable part. The banana skin is a part in the second
discourse but not the first discourse. There is no absolute or ebjectively given "part of
an banana Earnie, regardless of our sympathies as businessmen and advertisers, lied.
Rommetveit concluded that we must "concede that we do not (yet) know the ultimate,
objective woi Id and have hence to abandon the notion of any literal, basic, invariant and
"most-down-to-earth" meaning of banana" (p. :30).
I am. of course, reluctant to abandon the notion of the autonomy of meaning of
sentences. let alone words. So let us see if there is some way to meet Rommetveit's
criticism. First, I believe we may save the idea of the fixity of meaning the i.votcl
banana or of part by noticing that meanings are conventionalized symbols and are not
explained by appeal to the referent. Meaning and reference. as I said earlier. are
distinguished. Admittedly, the meaning of a term is based on one's knowledge of the
referent but also on one's knowledge of the other meanings in the language. The word
part, for example. has meaning relative to the mea .ng of the word whole, and so on So
one may think of the meaning of the word part without thinking about any particular
part or part of any particular object.
11
Rut I rtinrpriP to RnmmPrvpit that thp meaning of p^rt is not Incicpcndunt of
discourse. The meaning of part in biological discourse is different from its meaning in a
sharing discourse. Knowing the meaning depends upon knowing about that form of
discourse. Yet within that discourse. say biological discourse. part of an object again
has an autonomous meaning. Another example would be the meaning of proof in
mathematical as opposed to legal discourse.
Here we arrive at the critical point. The meaning of a word or of a piece of
discourse depends upon the properties of the discourse and not on the intentions of the
speaker. We cannot appeal to what Earnie meant when he said "I'll give you part of the
banana". We know exactly what Earnie meant. Eatme meant the skin. But the word
part in the discourse of sharink, means edible part if Rornmetveit is right and I believe he
is. It is that split between intention and meaning that makes interpretation an
interesting problem. Consequently. the central claim stands, the meaning of a word or
sentence is autonomous at least in the sense That it is independent of the intentions of
the speaker.
It is this point that we have sought to elucidate in our research. The question we
have asked is when do children come to distinguish speaker intention from sentence
meaning or as we sometimes put it, between speaker's meaning and sentence mean:ng.
And does their coming to make the distinction have any conceptual significance?
From the work of Newman and others we know that when they are around 8 years
old, the begin to note that, "technically" (I could say "literally"), Earnie did not lie.
They notice some gap between what is actunily said and what is intended. In our own
work we have been concerned with Lire beginnings of this understanding and its relation.
if any, to becoming literate.
In a typical study, Olson and Torrance (1986) children are told a story about Lucy,
Linus and Charlie Brown in which Lucy, wanting her new red shoes, asks Linus to bring
down her "red" shoes. The problem is that there are two pairs of red shoes in the closet.
12
Lanus. forced to guess. brings the wrong red shoes and Lucy. knowng Lucy. is not
grateful. The childien are asked a series of questions that pit their understanding of
Lucy's intention against the ambiguity of the utterance. The hypothesis is that young
children will not distinguish the intention from the meaning and will claim that Lucy
not only waited the new red shoes but also had asked for the new red shoes. Children
under six years of age, by and large. make this conflation.
Notice that their difficulty is not that they are literalists. Rather, the opposite.
They appear to b lieve that the sentence is an adequate representation of Lucy's
intention; they believe that what Lucy's sentence means is the same thing as what Lucy
means by it.
In the last two years. we have attempted to sort out mo, e precisely what is
involved here. There were several problems with the tasks we used. Perhaps children
did not remember exactly what was said. Indeed, we had frequent cases in which
children claimed that Lucy had said "new red shoes" when in fact she said just "red
shoes." And whereas we had interpreted this to mean that children conflated sentence
and intention others argued that they had simply forgotten the former.
Secondly, Robinson and Whittaker (1987) have recently argued that children's
handling of ambiguity could be explained without appealing the the speaker's intention.
It ma_ be that children simply agree to any utterance or accept any alternative
description as long as it is true of the object. Thus "red shoes" nd "new red shoes" are
both acceptable as they are both true of the refe -Ent. They write: ''If there is no relevant
discrepancy between the real wcrld oboe. t to v't ich a message refer- and the speaker's
internal representation of it. then we have no reason to assume that children considered
the speaker's internal representation." (p. 23 in 7" imeo version). Again: "Young
children.. ignore the.. speaker having an intention to communicate a particular
meaning." (p. 84 in published version.)
The problem is that an ambiguous expression like "red shoes" when there a/e two
13
pairs of red shoes is that the child may .ner think the utterance is about a particular
pair of red shoes and base its judgment on that or the child may think the utter ance is
about the speaker's internal representation of those particular shoes and ba-;e, its
judgment on that. What is needed is some method to pull those alternatives apart.
Perner. Leekam. and Yuille (in preparation) developed an ingenious task to do just
that. They did t by distinguishing the belief of the speaker from the object of the
utterance by instilling a false belief in the speaker. Children were told an enacted stow
in which a speaker v , vas mistaken in his belief about the location of an object. fold a
second person about the location by means of an ambiguous utterance. If the children
take the ambiguous utterance to 'oe a representation of the speaker's intention. they
should expect that the listener will come to share the speaker's false belief. If they
believe that the ambiguous message is about the intended object, then they should
expect the listener to think the utterance refers to the object's true location. Perner et al
found that 4 and 6 year old children assumed that the listener would intepret the
utterance to mean the location believed by the speaker rather than the true locatior
They conaded that young children identify the meaning of a message not with its
correct interpretation in context but with the intention of the speaker.
This result, however, has been criticized on the basis of the claim that perhaps
young subjects do not even notice the message and simply coral:se who believes what
about the location of the hidden object. For this reason Ruffman, Torrance and I
(Ruffman, Olson and Torrance. in preparation) have recently conducted a series of
studies on children's understanding of ambiguous utterances using Perner's false belief
senano. But first, let me introduce the general form of these experiments.
Children are introduced to sets of objects. typ cally two red boxes and one green
box and an object to be placed in one of the boxes by a puppet who plays the role of the
speaker. A second puppet plays the role of a listener It is about the listener's belief
that the child subject is asked. In a typical case, the first puppet places a candy in one
14
of the ,.ed boxes in the absence of the list -ler. The listener then enters the room and
asks the speaker where the candy is. The speaker replies with either an ambiguous
utterance "It's in the red box" or an unambiguous utterance "It's in the green box." The
child is then asked: "Does the listener puppet know where the candy is?" and "Where
does the listener think the candy is? Does he think it is in here (pointing), in here
(pointing) or does he think It could be in either?" The order of these questions was
counterbalanced.
In the first of these studies e found that children as young as 4 and 1 2 years
correctly claimed that the listener would know the object's location when the message
was unambiguous and would not know when the message was ambiguous. Below this
age, children thought that the listener would know waere it was and that the listener
would think it was where it really way. This understanding of ambiguity at such an
early age was achieved by frequently repeating the critical information namely that the
speaker said "red box" and "there are two red boxes." Nonetheless with such a task it
appears that children understand the consequences of ambiguity at least two yea's
earlier than our earlier research had suggested.
But the results to the "think" question are unexpected. Even after insisting that
the listener will not know where the candy is, when asked where he will think it is, none
of the children say that he thinks it could be in either red box. About half of the
children point to the correct red box, the other half to the incorrect rec' box. There is
little indication that they believe that the speaker's belief or the true location would "get
through" to the listener Moreover. even if a child were to go for the correct location it
would be unclear whether that child was picking the true location because it was the
true location or because the speaker believed it to be there.
Our most recent study, therefore, adopted Perner, Leekham and Yuille's procedure
of establishing a false belief in the speaker before the speaker attempts to communicate
that belief to the listener by means of an ambiguous message. In this way the truth of
15
k t cnunlenr n Tk nthe mess',ge ...e ,ueso".is wil_ the child think that the listener will come to acquire the speaker's false belief
even if the message is ambrguous?
To this end, a sEnario was constructed in which the speaker hides a candy in one
of the red boxes and then teases the scene. Children are reminded that the listener did
not see him put the candy in a box. Next. with neither the speaker nor the prospective
listener present, the child and the Experimenter move the candy to the other red box.
Again children are reminded that the speaker and the listener do not know that. Now
the speaker and the lis'erer dolls appear and the speaker tells the listener doll that the
candy is in the "red box." an ambiguous message. The child is again reminded of
exactly what happened and who knows what. Then the child is asked a series of control
questions to determine that they remember the details. and then is asked the following
critical questions:
Does the listener know where it is?
Where does the 113teler think it is. in here (pointing to one red box), in here(pointing to the other red box) or does he think it could be Ln either?
Why does he think its in there?
First, the know question. As in the preceeding study, almost all of the children
correctly say that the listener will not know. Notice that they may be saying the listener
does not know for either of two reasons. They may, as in the prec-ding study, say he
does not know because the speaker's message was ambiguous. Or they may be saying he
does not know because he now shares the false belief with the speaker. Responses to the
"think" question help to sort this out.
Preliminary results for the "think" question are presented in Table 1 for t;vo
separate runs of this experiment (in the second run, Experiment 2, children were also
asked the question "Where does the speaker think it is? All subjects answered that
question correctiy). These data make clear that there Ls an overwhelming tendency for
10
16
the children to claim that the listener will think the candy is in the box that the speaker
believed it to be in. This suggests that the reason they answer the "know" question the
way that they do is that they genuinely believe that the listener has come to hold a false
belief. They appear to have ignored the ambiguity altogether when the speaker holds a
false belief. Yet when the listener holds a true belief as in the preceding experimen!,
they recognize the consequences of ambiguity.
We are currently collecting data from children in a variety of control conditions.
To date we are quite certain that if the speaker's belief is true 5 year olds claim that a
listener will know the location if the utterance is unambiguous but not if it is
ambiguous. Hence, we know they pay attention to the message. But we also know that
if the speaker's belief is false, children believe that the speaker's intention (his belief)
will be conveyed even by an ambiguous message Just as it will by an unambiguous
message. Ruffman has proposed that one final control is needed, one which pits the
speaker's belief against the meaning of .iis message, an arrangement that can be
achieved by the speaker lying, and then see what the child infers the listener believes.
We expect that the message. If unambiguous, will override the speaker's beliefs.
What do we infer from all this? We believe we have established that children into
the school years believe that speaker's intentions, especially when the speaker's belief is
false, make their way to the listener even when there is no lexical means for conveying
that intention. We take this as evidence for our original h, pothesis. namely, that
children tend to conflate what a speaker means with what the utterance means. With
age and schooling they come to see these two "meanings" as independent.
Another recent study (Torrance. Howes and Olson) suggests that if the speaker's
message is in the form of a written text, children are much more 11k ly to notice the
ambiguity of the utterance and acknowledge that the listener will have no grounds for
any particular belief. Thus, literacy appears to be instrumental in sorting out this
distinction although much more work is needed before this conclusion can be stated with
I -I
any certainty.
Children's discovery of the autonomy of sentence meaning is critical for a variety
of educational tasks. Revision in writing is a matter of bringing the two "meanings"
into congruity. When reading shifts from mere assimilation to thinking, it is because of
the same factor. Children, in this case, come to revise their estimate of the meaning
intention on the basis of the linguistic or sentence meaning. The begin to treat that
textual meaning as autonomous.
Now, finally, is textual meaning really autonomous? I would acknowledge that it
is not. Texts are always open to re-interpretation as Nystrand has argued in his
discussion )f legal texts. Not only do their meanings change as contexts change but also
the textual or sentence meanings change as cultural conwntions change. So there is ao
absolute meaning of a text. Nor is there one true intention of which a text is a
fragmentary expression. But that is the limit of my concession to critics of "autonomy"
of text. What continues to seem to me to be the case is that the basic distinctions
required for literate interpretation continue to be useful for distinguishing between the
properties of the text which are "taken as given" for any particular purpose and the set
of construals or interpretation:, that can be made of that text. The distinction is exactly
that between facts and theories. Most of us agree that there is no absolute distinction
between facts and theories; today's fact turns out to hide a host of theoretical
assumptions. Nonetheless, theoretical science proceeds by "taking as given" a set of
facts which can be organized relative to some theoretical scheme. In learning to
distinguish meanings from intentions, and thoughts from expressions, children are
taking the first giant step in that direction.
18
References
Austin, J. L. (1965) How to do things with words. New York: Oxford University Press.
Anderson, R. (1977). The notion of schemata and the educational enterprise. InR. Anderson. R. Spiro & W. Montague (Eds,) Schooling and the acquisition ofknowledge. Hillsdale, NJ: Lawrence Erlbaum Associates.
Boring, E. G. (1950). A history of experimental psychology. New York: Appleton-Cen tury-Crofts.
Feldman, C. (In press). Oral metalanguage. In D. Olson & N. Torrance (Eds.) Literacyand orality. Cambridge: Cambridge University Press.
Finnegan, R. (Ed.). (1978). The Penguin Book of African poetry. London: Penguin.
'rancis, H. (1987). Cognitive implications of learning to read. In Olson, D. (Ed.)"Understanding literacy." Special issue of Interchange, 18, 1-2, pp. 1-171.
Frege, G. (1952). Sense and reference. In P. Geach & M. Black (Eds.) Philosophicalwritings of Gottlob Frege. Oxford: Blackwell.
Gadamar, H.-G. (1975). Truth and method. London: Sheed and Ward.
Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: EiougtitonMifflin.
Goody, J. & Watt, I. (1963). The consequences of literacy. Contemporary studies tosociety and history, 5, 304-45. Republished in J. Goody, (Ed.) (1968). Literacy inTraditional Societies. Cambridge: Cambridge University Press.
Goody, J. (1987). The interface between the oral and the written. Cambridge: CambridgeUniversity Press.
Hansen, R. (1958). Patterns of discovery. Cambridge: Cambridge Press.
Harriman, M. (1986). Metalingutstic awareness and the growth of literacy. In S. deCastell, A. Luke & K. Egan (Eds.), Literary, society, and schooling. Cambridge.Cambridge University Press.
Luria, A. R. (1976). Cognitive development: Its cultural and social foundations.Cambridge, Mass.: Harvard University Press.
McLuhan, M. (196.). The Gutenberg Galaxy. Toronto: University of Toronto Press.
Narasimhan, R. (In press). Literacy: Its characterization and implications. In D. Olson& N. Torrance (Eds.) Literacy and orality. Cambridge: Cambridge UniversityPress.
Newman, D. (1982). Perspective taking versus content in understanding lies. Thequarterly Newsletter of the Laboratory of Comparative Human Cognition, 4, 26-29.
Nystrand, M. (1986). The structure of written communication: Studies in reciprocitybetween writers and readers. Orlando, FL: Academic Press.
Olson, D. (1970). Language and thought: Aspects of a cognitive theory of semanticsPsychological Review, 77, 257-273.
Olson, D. (1977). From utterance to text: The bias of language in sper.C,ng and writing.Harvard Educational Review, 47, 254-279.
Clson, D. (Ed.) (1987). Understanding literacy. Special issue Interchange. 18, 1-2.
19
Cilcnn 11. R. Tnrranra N (1_ QAT) I iterary and rtignitivP riPvPlopmont A ,nryseptuaitransformation in the early school years. In S. Meadows (Ed.), Developingthinking: Approaches to children's cognitive development (pp 142-160).London: Methuen.
Olson, D. & Astrigton, J. (1989) Talking about text: How literacy contributes tothought. Journal of Pragmatics.
Perner, J., Leel,ham, S., & Yuille. N. (in preparation). The message says what thespeaker means.
Popper, K R. (1972). Objective knowledge. Oxford: Oxford University Press.
Robinson, E. & Whittaker, S. (1987). Children's conceptions of relations betweenmessages. meanings and reality. British Journal of Developmental Psychology, 5,81-90.
Rommetveit, R. (1988) On literacy and the myth of literal meaning. In R. Salo (Ed.),The written world: Studies In literate thought and action_ New York: Spcinger-Verlag.
Ruffman, T., Torrance, N., & Olson, D. R. (in preparation). Children's understanding ofambiguity.
Scribner, S. & Cole, M. (19F11). The psychology of literacy. Cambridge, Mass.: HarvardUniversity Press.
Smith, F. (1971). Understanding reading. Toronto: Holt, Rinehart and Winston. Inc.
Street, B. (1984). Literacy in theory and practice. Cambridge: Cambridge UniversityPress.
Torrance, N. & Olson, D. (1989). Children's understanding of ambiguity andinterpretation. San Francisco: AERA.
Table i
Children's Ascriptions of Beliefs to rile Listener Doll in False Belief - Ambli:thiu,Condmon
4
Age
5 6 Total
Experiment 1:
Intended 5 11 18 34
Actual 1, 3 6 10
Either 0 2 I 3
Experiment. 2:
Intended 4 15 19
Actual 0 3 3
Either 0 0 0
Do they go by what the speaker meant
said
believed